Your undergrad and masters research journey: FAQs
In the year 2019, I first began working with a team on a research project. That was my first experience as a researcher, and it still is a roller coaster ride. I have come across juniors who often want to get into undergraduate/graduate-level research but are often lost. In this blog, I would like to answer some common questions people often ask when beginning their undergraduate or master’s level research.
I am a 2nd year BTech/1st year MTech student; how do I choose a domain?
When you have joined the course, you will have an idea of the various domains of research. I can speak for computer science; the most popular research areas are AI/ML, cybersecurity, blockchain technologies, IoT, etc. You can try reading blogs or check out YouTube videos (such as Yannic Kilcher’s channel for ML) to get an idea of the work done in these domains. Another great source to understand the current research trends in an area is to go to Google Scholar and read survey papers into a specific domain. Survey papers discuss the recent developments into one big research problem and often compare the results, which is an excellent resource for understanding the ground.
I know my domain. What are the first things I would need to do?
This confuses a lot of students (myself included). We just don’t know how to take the first baby steps into a new world. An easy way to start is to look at the profiles of the faculties in your university. You need to check their research domains and see whose domain interest you the most. Pick your niche, and then begin by reading up their most recent research papers. A quick look at their website or Google scholar page would give you a list of their publications. This read-up will provide you with an insight into what kind of research he is interested in. Once you’ve identified your domain, approach the professor.
Bonus point: You can also talk to the Ph.D. candidates and Postdocs in your university to understand their research.
What is the process of “doing research” is like?
A research project aims to solve unsolved problems or propose better solutions over existing solutions. How can we build algorithms to detect cancer from MRI images? How can a computer translate a text from one language to another? How can we make bio-chips to automatically monitor health parameters on the go?
The basic workflow of “doing research” is quite simple. First, perform a literature survey. Read papers that are written on the topic you’re working on. This literature survey will give you an idea of the previous researchers’ techniques and the problems in those approaches.
Secondly, armed with the knowledge of previous work, you might design a new method of solving the problem. In this part, you will be heavily assisted by Ph.D. candidates, senior master’s students, and postdocs.
The third step includes performing experiments on the method. You would conduct lots of controlled experiments on your design and comparing and analyzing the results. This step might be the most time-consuming one. This might also be quite discouraging because your investigations will not give the desired results more often than not. You mustn’t lose hope and continue.
Final step. Did you get good results? Did you build a novel solution that has a lot of potentials? Excellent, write a research paper to publish your findings.
What is a literature survey, and how do I do it?
A very crucial part of your research journey. A literature survey is a process of reading research papers published previously in a research domain. Your best friend in this regard would be Google Scholar, where you can type in the search words and read the papers. You’re recommended to read documents with good citation numbers. Citation number indicates how many people referred to the work. Bigger the citation number, the better the quality of the paper. It is also recommended to refer to the most recent documents (no more than 5 years old). This is partly because the research field in computer science (especially in ML) is changing at a scary speed. Some would even consider 6–7-year-old research to be practically pre-historic.
